Reproductive Biology and Breeding Systems of Two Opisthopappus Endemic and Endangered Species on the Taihang Mountains

Opisthopappus is a perennial, endemic herb of the Taihang Mountains in China. Two species of this genus (O. longilobus and O. taihangensis) are important wild genetic resources for Asteraceae; however, their reproductive biology has been lacking until now. This study is the first detailed report on the reproductive biology and breeding systems of two Opisthopappus species. Through field observations, the floral syndromes of O. longilobus and O. taihangensis were found to possess a similar pattern, although O. taihangensis has a relatively larger capitulum, more ray ligules, and disc florets. The flowers of both O. longilobus and O. taihangensis are protandrous, a character that can prevent autogamy at the single-flower level, and insects are required for pollination. Further, brightly ligules, brightly bisexual florets, unique fragrance, and amount of nectar suggest that these species propagate via an entomophilous pollination system. Hymenopteran and Diptera species were observed as the effective pollinators for these two species. The outcrossing index, pollen/ovule ratio and the results of hand pollination indicated that these Opisthopappus species might have a mixed mating system that combines cross-fertilization and partial self-fertilization for O. longilobus and O. taihangensis, outcrossing predominated in the breeding system, while self-pollination played an important role in seed production when insect pollination was unavailable, particularly in a harsh environment, such as the Taihang Mountains cliffs. Meanwhile, O. taihangensis might better adapt to severe surroundings with relatively complex floral syndromes, specifically through the attraction of visiting insects and a high seed set rate. The above results not only provide reference information toward a better understanding of the survival strategies of O. longilobus and O. taihangensis in the Taihang Mountains but also lay a solid foundation for further exploring the molecular mechanisms that underly their adaptation under cliff environments.


Introduction
Reproduction is not only the most important and relatively fragile step in the life cycles of plants but also the core of their evolutionary process [1][2][3][4]. Meanwhile, pollination biology is a critical parameter in the life history of flowering plants that determines the success of sexual reproduction [5][6][7]. The mating and breeding systems of plant species are determinant factors of the genetic structures of populations [7,8]. Thus, an investigation of the characteristics of the reproductive biology of a species, including their pollination ecology and breeding system, is indispensable for elucidating the different phenophases of a species. Further, to explore the strategies they employ to adapt to changing habitats, particularly for species that thrive in harsh environments [4,6].
The flowers bloom from mid-September to late October and peak in early October seeds mature at the end of November.
The flowering time of the two Opisthopappus species was ~60 days, and a s flower would unfold for ~seven days. Interestingly, purple or green flower buds observed for both O. taihangensis and O. longilobus. For the green buds, the ligules taihangensis and O. longilobus were always white from unfolding to end. For the pu buds, the ligules of O. taihangensis were general purple or lilac at first and grad changed to light pink or white until fully unfolding, while the ligules of O. longi were pale pink or white with unfolding ( Figure A1).
The floral phenology of a single capitulum could be divided into nine stage bud stage, (2) pre-bud break, (3) late bud break (bud breaks and the internal structu the flower not present), (4) initial unfolding stage (ligules initially unfold and the i nal structure begins to present), (5) ligules unfolding stage (ligules fully unfolding florets fully exhibiting), (6) florets begin to unfold (the outermost layer of the disc fl begin to unfold, pistils mature, and stamens begin to disperse pollen), (7) florets unfolding (the innermost layer of the disc florets completely unfolding), (8) floral d (the ray ligules and disc florets begin to wither), (9) complete floral defeat (the flo completely wither), and (10) seed maturity (seeds maturation).
Based on field observations, the developmental stages of capitulum maturatio O. longilobus and O. taihangensis were really the same without obvious differences ure 1A,B). The capitulum dimensions of O. taihangensis were slightly larger than those of O. longilobus (p < 0.05). The numbers of ray ligules and disc florets for O. taihangensis were more than those of O. longilobus (p < 0.05). These showed that O. taihangensis had a relatively larger capitulum and more numbers of ligules and florets. The detailed floral morphological characteristics are presented in Table 1. Note: the first value was the mean of flower traits; the second value was the standard deviation (SD). a, nosignificant variation; b, significant variation; p< 0.05.
The capitulum of both O. longilobus and O. taihangensis were comprised of ray ligules and disc florets, where the external ligules were all female with stigmas, while the internal florets were all bisexual flowers that simultaneously contained stamens and pistils ( Figure S1). In addition, it was found that most O. longilobus and O. taihangensis stigmas were two-lobed, while there were only a few that were three-lobed ( Figure S1). Thus, the floral organ structures were not significantly differentiated between the two species.
All the florets of O. longilobus and O. taihangensis were hermaphrodites. The corolla of the florets was sympetalous with a five-lobed apex. The filaments of stamen attached to the edge of the corolla, as well as the anthers, were connected to form a five-lobed tube around the style. According to the stamens of O. longilobus and O. taihangensis, both were polygamous.
At the onset of flowering, both the stamens and pistils were enclosed within the petals of the florets, which opened upon blooming to expose the filaments with anthers that were gradually raised, whereafter the pollen sacs began to rupture and disperse pollen. Meanwhile, the stigmas began to elongate and become longer than the stamens, and steadily the penniform spread out ( Figure S2). As relates to the internal florets, the external ones initially pinnated and dispersed pollen, subsequently followed by the internal florets ( Figure S2).
]The O. longilobus and O. taihangensis stigmas were initially short and rod-shaped, which soon transformed to a "Y"-shape over time and finally to a curly and oxhorn-shape ( Figure 2). Scanning electron microscopy revealed that the stigmas were composed of finger-like cells at the top and protruding cells on both sides, which could secrete mucus and adhere to the pollen. Thus, the pollen was successfully deposited on the finger-like cells and protruding cells of the stigma and sprouted pollen tubes ( Figure 2).

Floral Visitors
The behaviors of potential pollinators were significantly influenced by external vironmental factors (e.g., temperature) [25]. Field observations revealed that the flow visiting behaviors of insects occurred mostly on sunny days and showed a single-pea curve of the daily activity pattern.
The activities of flower-visiting insects occurred mainly from 11:00-14:00, with m insects visiting from 12:00 to 13:00. On sunny days, visiting insects were rarely obser when the temperature was lower (8:00-11:00). Thereafter, the number of visiting inse decreased rapidly due to lower temperatures. Almost no flower-visiting insects were served after 15:00 on sunny days or on cloudy/rainy days. According to comma obser tions, the number of O. taihangensis flower-visiting insects was higher than that for longilobus during the times described above (Table S1).

Floral Visitors
The behaviors of potential pollinators were significantly influenced by external environmental factors (e.g., temperature) [25]. Field observations revealed that the flowervisiting behaviors of insects occurred mostly on sunny days and showed a single-peaked curve of the daily activity pattern.
The activities of flower-visiting insects occurred mainly from 11:00-14:00, with most insects visiting from 12:00 to 13:00. On sunny days, visiting insects were rarely observed when the temperature was lower (8:00-11:00). Thereafter, the number of visiting insects decreased rapidly due to lower temperatures. Almost no flower-visiting insects were observed after 15:00 on sunny days or on cloudy/rainy days. According to comma observations, the number of O. taihangensis flower-visiting insects was higher than that for O. longilobus during the times described above (Table S1).
Further, Scathophaga stercoraria (Diptera), Oxycetonia jucunda Faldermann (Coleoptera), Vespa velutina (Hymenoptera), and the Parasitoid wasp (Hymenoptera) typically visited O. longilobus and O. taihangensis flowers. Diptera and Hymenoptera, which may also be effective pollinators, were observed landing on the corolla; however, their visiting frequency was relatively low. Oxycetonia jucunda Faldermann (an insect of Cetoniidae (Coleoptera)), which also alighted on O. longilobus and O. taihangensis flowers, albeit with the least visiting frequency and remaining on flowers for only a few seconds, may be a pollen thief with its proboscis.
Further, Scathophaga stercoraria (Diptera), Oxycetonia jucunda Faldermann (Coleoptera), Vespa velutina (Hymenoptera), and the Parasitoid wasp (Hymenoptera) typically visited O. longilobus and O. taihangensis flowers. Diptera and Hymenoptera, which may also be effective pollinators, were observed landing on the corolla; however, their visiting frequency was relatively low. Oxycetonia jucunda Faldermann (an insect of Cetoniidae (Coleoptera)), which also alighted on O. longilobus and O. taihangensis flowers, albeit with the least visiting frequency and remaining on flowers for only a few seconds, may be a pollen thief with its proboscis.

Pollen Viability and Stigma Receptivity
A pollen viability test was conducted at different flowering stages. On the first, second, and third days of anthesis, the pollen of O. taihangensis had the highest viability with a mean value of 92.12 ± 0.12% (Table 2). On the fourth day, the pollen viability began to decrease to 80.4 ± 0.47% and continually decreased on the fifth, sixth, and seventh days.
These data suggested at least a seven-day lifespan for the pollen grains of O. taihangensis and O. longilobus. Meanwhile, the pollen viability of O. taihangensis was superior to that of O. longilobus (p < 0.05).
The change trends for O. taihangensis and O. longilobus pollen viability were similar on the same flower unfolding day, as they both displayed a single peak ( Table 2). The pollen viability of the two species gradually increased from 8:00 to 13:00 and then decreased from 13:00 to 17:00. The highest pollen viability was found at 13:00 for both O. taihangensis and O. longilobus. Moreover, the viable pollen number of O. taihangensis was higher than that of O. longilobus during the same times (p < 0.05).
The receptivity of the stigmas was found to gradually increase during the first 1-4 days of the flowering period (Table 3, Figure 4). For O. taihangensis, they exhibited the highest receptivity on the third and fourth days, while the highest level of O. longilobus stigma receptivity occurred on the fourth day. Subsequently, the stigma receptivity for the two species began to decrease on the fifth and sixth days, then decreased to zero on the seventh day (Table 3, Figure 4). Note: +, presenting the receptivity; ++, a stronger receptivity; +++, the strongest receptivity; very weak receptivity or sterility. The pollens were sampled at 8:00-17:00 for 1, 2, 3, 4, 5, 6, a days to observe. On the same day, during the flowering phase, the stigma receptivity of O. taihan sis and O. longilobus exhibited a similar change pattern (Table 3, Figure 4). The stigma the two species showed the highest receptivity level at 11:00-13:00, around which ab dant bubbles were emitted. Subsequently, the stigma receptivity decreased with fe bubbles over time.
At a different stage, it was observed that the closed stigmas with few bubbles ex ited weak receptivity. When the stigmas opened and appeared in their "Y" config tion, their receptivity increased around more bubbles. When the stigmas appeared rolled Y shape, the receptivity displayed the highest level around ample bubbles (Fig  4). On the same day, during the flowering phase, the stigma receptivity of O. taihangensis and O. longilobus exhibited a similar change pattern (Table 3, Figure 4). The stigmas of the two species showed the highest receptivity level at 11:00-13:00, around which abundant bubbles were emitted. Subsequently, the stigma receptivity decreased with fewer bubbles over time.
At a different stage, it was observed that the closed stigmas with few bubbles exhibited weak receptivity. When the stigmas opened and appeared in their "Y" configuration, their receptivity increased around more bubbles. When the stigmas appeared as rolled Y shape, the receptivity displayed the highest level around ample bubbles (Figure 4).

Outcrossing Index and Pollen/Ovule Ratio
The average diameters of the floret's inflorescence (bearing part of the capitulum) of O. taihangensis and O. longilobus were 5.68 ± 0.12 cm and 4.55 ± 1.22 cm, respectively. Both of these diameters were longer than 0.6 cm and thus were scored as 3 according to the Dafni criterion [26]. When the florets of O. taihangensis and O. longilobus unfolded, the anthers in the same capitulum could dehisce and disperse pollen during the stigma receptivity period, although having protandry. This indicated that completely self-reproductive isolation did not occur in O. taihangensis or O. longilobus. Rather, a degree of self-compatibility occurred in the two species; thus, they were scored as 0. Meanwhile, the stigmas were longer than the anthers for O. taihangensis and O. longilobus (p < 0.05), which formed a certain degree of spatial isolation with a score of 1.
According to Dafni's criterion, the outcrossing index (OCI) for both O. taihangensis and O. longilobus was considered to be 4. This revealed that the reproductive system should be  [27], the breeding system of both species was facultative xenogamy, which was consistent with the OCI classification.

Seed Set and Germination Rates
Without bagging and artificial pollination, the seed rates were 40 Under different treatments, the seed set rate values generally exceeded 20%, which further suggested that O. taihangensis and O. longilobus were self-compatible to a certain extent.
According to the above different treatments, the mature seeds were respectively collected at different sites from 2018 to 2019, and seed germination was subsequently performed at room temperature conditions in the laboratory. Then, we found that the average seed germination rates were 53.33 ± 2.01% for O. taihangensis and 43.33 ±1.47% for O. longilobus. Meanwhile, the seed germination potentials of O. taihangensis and O. longilobus were 36.00 ± 0.28% and 25.00 ± 1.02%, respectively.
According to the above, the seed germination rate and potential of O. taihangensis were higher than that of O. longilobus (p< 0.05), even at the same sampling site.

Floral Syndromes and Effective Pollination
Floral syndromes comprise sets of integrated floral traits (e.g., morphology, color, and scent) [4,25,28]. For the two species of Opisthopappus, the floral syndromes were similar ( Figure 1) except for the capitulum size and the number of ligules and florets ( Table 1). As the descendant of O. longilobus [19], O. taihangensis presented floral syndromes that were similar to those of its ancestors. On the other hand, the relatively complicated floral syndromes of O. taihangensis might be an evolutionary outcome. With a larger capitulum, more ligules, and florets, O. taihangensis could better attract pollinators for reproductive success, development, and survival when it faces different habitats compared with O. longilobus.
For plants, floral traits are associated with different pollination syndromes [4], which may evolve to align with the morphologies and behaviors of pollinators [29]. The floral syndromes of O. taihangensis and O. longilobus (e.g., bisexual disc florets, brightly white ligules, brightly yellow florets, and protandry) indicated that their pollination systems should be animal-mediated cross-pollination. In the lower portion of the florets, nectary secretory tissue lined and generated a certain amount of nectar. Nectar can release a fragrance and provide precious forage for flower-visiting insects, particularly in midautumn. The capitula inflorescences with receptacles are beneficial for insect visitors to remain for prolonged periods. Thus, the results of flower-visiting observation pollination support that the pollination system of O. taihangensis and O. longilobus should be animalmediated cross-pollination.
The main floral visitors of O. taihangensis and O. longilobus were found to be hymenopteran and Diptera species (Figure 3). Our results confirmed that they shared the same flower-visiting insects, as they possess similar floral morphologies (Figures S1 and S2). Meanwhile, the flowers have characteristics that are associated with bee and fly pollination (Figure 3), such as small tubular corollas and sites with concentrated nectar. Bees generally have pollen-carrying legs that are completely brosse [7,30]. Visiting bees can easily carry pollen when landing on the florets and transport it from one flower to the next. Since the bodies of flies are generally smaller than those bees, they may enter the corolla tube to feed on nectar and then repeatedly contact the stigma of different flowers during foraging [30,31]. Thus, the data suggested that bees and flies (e.g., Vespula vulgaris and Episyrphus balteatus) were the effective pollinators of O. taihangensis and O. longilobus (Figure 3, Tables S1 and S2).
Interestingly, O. taihangensis and O. longilobus were observed to have unique flowervisiting insects (Tables S1 and S2). Two Hymenoptera species (Vespa velutina and Parasitoid wasp) were specific for O. taihangensis, which can not only land on but also enter the florets with their slender bodies. Meanwhile, two Lepidoptera species, Aricia mandschurica Staudinger and Polygonia caureum were specific for O. longilobus (Figure 3). The butterflies have siphoning mouthparts and can suck nectar through their long tubular beaks. When they visit flowers, their body hairs can adhere to and transfer pollen [31,32]. The differentiation of the flower-visiting insects between O. taihangensis and O. longilobus may reflect the floral syndrome variations, such as the flower size and color, particularly the flower fragrances, where different aromas might attract various insects. Metabolome data revealed that significantly different floral metabolites were found between O. taihangensis and O. longilobus [33].
The extended floral longevity of these plants may be an evolutionary strategy that is employed to prevail over sparse or unpredictable pollinator services by increasing the quantities of their own exported pollen and imported foreign pollen [4,34]. Our results revealed that the flowering duration of the two Opisthopappus species extended for about 60 days, while the anthesis of a single flower continued for six days. The pollen release time was relatively concentrated. The pollen of O. taihangensis and O. longilobus gradually matured from the external to internal florets (Table 2, Figure S3), with blossoms being relatively concentrated in October. The extended flowering time and pollen presentation could increase the time of pollen collection for pollinators [35]. Furthermore, the extended floral longevity was likely an adaptation for survival and development in the difficult cliff environment of the Taihang Mountains.
For most flowering plants, pollen viability and stigma receptivity are critical parameters for the successful initiation of pollen-pistil interactions [7,36]. Extended stigma receptivity assures the pollination, fertilization, and reproductive success of plants [14]. The stigma of O. taihangensis and O. longilobus can remain receptive for seven days to provide the proper conditions for pollen to germinate and affect fertilization (Table 3, Figure 4). This time frame appeared to be a strategy for ensuring the reproductive output of the two species [7]. The anther dehiscence in O. taihangensis and O. longilobus occurred simultaneously with anthesis. Alternate anther dehiscence appeared to be an adaptation for efficient and extended pollen availability to the pistil, which may further increase the chances of being pollinated [4,7].
Despite successful insect pollination, the activities of pollinators can be easily influenced by weather conditions, as in this study, almost no visiting insects were observed on rainy or cloudy days. We found that the flowering periods of the two Opisthopappus species of sampled sites basically overlapped with the rainy season (November to October), which had a negative impact on the flower-visiting behaviors of insects. Under this situation, successful reproduction may rely solely on self-pollination or geitonogamy [18].
Insect pollinators preferred warmer noon hours to visit the flowers, after which their visitation frequency decreased (Tables S1 and S2) [7,37]. Different plant species experienced a drastic decrease in the visitation efficiency of pollinators during windy and cloudy periods and rain [38]. Adverse climatic conditions were found to hamper the visitation frequency and efficacy of the pollinators [18]. Lower anthesis temperatures were unfavorable for the pollinators [7]. The low temperatures of early autumn in North China may also decrease the diversity and activities of pollinators [39], especially in the cliff habitats of the Taihang Mountains [4,40]. However, the long anthesis period of each flower, as well as the capitulum and light flower color, with shared pollinators, still improved the pollination efficiency and resulted in the mass production of seeds for O. taihangensis and O. longilobus.

Breeding System
The breeding systems of plants are a manifestation of the interactions between their internal genetic mechanisms and the external environment, which plays an important role in the evolution and variation of species [41]. Knowledge regarding the breeding system of a species is beneficial for elucidating the characteristics of evolution and life history caused by different genetic and ecological factors [6].
It is widely known that the outcrossed progeny of organisms are expected to be more heterozygous; hence, individually more adaptive and with more fitness than selfed progeny [18,42]. However, in flowering plants, pure outcrossing (e.g., self-incompatible) may bring potential risks to the number of offspring, as cross-fertilization relies on an external agent for pollen transfer (e.g., animals and wind), which is often unpredictable and/or inadequate [6]. Thus, in nature, there are very few plants that are exclusively self-crossing or out-crossing. Instead, most modern plant species evolved toward a mixed mating system combining self-fertilization and cross-fertilization for reproductive success [4,41], particularly as a strategy to cope with unstable environments [4,43].
According to Dafni [26], the out-crossing index (OCI) was 4 for this study, which indicated that these Opisthopappus species were self-compatible. Further, the pollen-ovule ratio (P/O) was in the range of 244.7-2588.0, which corresponded to the values reported for facultative xenogamy [43]. Hence, O. taihangensis and O. longilobus were a mixture of self-pollination and out-crossing. The mixed breeding strategy ensures reproductive output that not only ensures sexual reproduction but also generates variation and distributes effectively [7].
The evolution of plant mating strategies always has one main theme related to the evolution of cross-vs. self-fertilization [44,45]. Predominant selfing and outcrossing should be alternative stable outcomes of mating system evolution for most plants [25,46]. Our results revealed that O. taihangensis and O. longilobus could propagate via simultaneous selfing and outcrossing; however, outcrossing predominated. The seed set from geitonogamous, xenogamy, and control was higher, respectively, than that from autogamy. The seed germination rate from autogamy was lower than that from geitonogamous, xenogamy, and control, which further supported that outcrossing prevailed in the breeding system. Selfpollination played a secondary role in ensuring production in the breeding system when the conditions for outcrossing were unfavorable due to shifting environment conditions, or in cases when normal pollinator populations were reduced or absent [25].
Dichogamy and herkogamy are the two main mechanisms that prevent spontaneous self-fertilization [4,47]. Although O. taihangensis and O. longilobus both displayed protandry, the seed rate of manual self-pollination exceeded 20%, which indicated that spontaneous selfing occurred. These two species exhibited several floral traits that were specifically associated with self-pollination (bisexual florets and no temporal separation between pollen viability and stigma receptivity) [48]. However, the fruit set and germination rate from autogamy were significantly lower than that from xenogamy and control.
Inbreeding/selfing is relatively common in species with small populations and individuals, narrow habitats, or modest founding populations [6,49,50] and is of great significance for their preservation and reproduction. Both O. taihangensis and O. longilobus grow on cliffs of the Taihang Mountains and thus are subjected to a harsh environment. Sustained self-compatibility would be a strategy during the evolutionary process, as a delayed mechanism for reproductive assurance [51,52], or even as a main contributor to natural seed setting [18].
Moreover, selfing results from the activities of pollinators. The numbers of flowers unfolding simultaneously are important for the reproductive success of species that are dependent on biotic vectors since this can increase pollinators' attraction and visitation [53,54]. However, visits will frequently result in geitonogamous self-pollination [4,55]. Thus, the high fruit set rate under ambient conditions for O. taihangensis and O. longilobus is likely the result of both cross-pollination and geitonogamy.

Study Species and Sites
O. longilobus and O. taihangensis are autumn flowering plants which possess pinnatifid leaves and heterogamous capitulum. Their ray ligules are general lamina linear, white, or pink. The disk florets are bisexual, yellow, and five-lobed. The fruits and seeds are a type of obovoid achene, 1.2 mm in length, with three or four ribs [56].
O. longilobus and O. taihangensis typically grow on steep cliffs at altitudes of from 1000-2000 m, along with other coexisting plants (some Gramineae) in wild habitats. Due to the harsh environment, they were difficult for us to observe. Thus, we created a common garden in Linfen City (CLF) of Shanxi Province, along with transplant gardens at Shennongshan (SNS) in Henan Province and Xiangtangshan (XTS) of Hebei Province. Our fieldwork was conducted at these three sites from July to November.
The CLF site is located at the Shanxi Normal University Nursery (36 • 60 N, 111 • 30 E) in Shanxi, China. The climate of the CLF belongs to a type of temperate continental monsoon, which is characterized by cold winters with little snow, spring winds, autumn rains, and hot summers with droughts. The average annual temperature is 12.6 • C, with 500-600 mm of precipitation and a frost-free period of 190 days [57]. The SNS is set in the Shennong Mountains (35 • 11 30" N-35 • The climate here is warm-temperate continental with four distinct seasons, which are dry and windy in spring, hot and rainy in summer, with warm days and cool nights in the autumn, and a cold and dry winter. The average annual temperature is 14.3 • C, with maximum/minimum temperatures of 42.1 • C and −18.6 • C, respectively [58]. The XTS is located in the Xiangtang Mountains (36 • 20 N-36 • 20 N, 114 • 03 E-114 • 16 E), in Hebei, China, with a warm temperate continental monsoon climate. This site also has four distinct seasons, with an average annual temperature of 13.5 • C and an annual frost-free period of 200 days [59].

Pollinator Observations and Pollination Experiments
Insect observations were conducted continuously during peak flowering in the field, from 10:00 to 18:00 for 60 d in 2018 and 2019, from September to October. Each insect was placed in a separate vial, where the presence or absence of pollen grains adhered to their bodies was determined in the laboratory using a stereomicroscope. The species were identified by associate professor Junxiang Cai (an insectologist at Shanxi Normal University). The insect specimens were preserved in the specimen room of the School of Life Science at Shanxi Normal University.
The times and frequencies of insect visitors to O. longilobus and O. taihangensis flowers were recorded over 30 consecutive days, which totaled 500 flowers under observation for each species. The pollinators and their foraging behaviors were analyzed by direct visual observations, which were complemented by photographs taken during visits to the three sites. Based on their visitation frequency and behavior on the flowers, the insects were classified as effective pollinators or occasional pollinators. Effective pollinators represented the frequent insects (with a higher number of visits) that carried pollen grains from one anther of a flower to the stigma of another. Occasional pollinators represented those insects that showed the same behavior but with a low visitation frequency. Insects that only collected pollen and/or nectar without pollinating the flowers were classified as thieves [60,61].

Pollen Viability and Stigma Receptivity
After flowering, ten flowers were sampled daily at 1, 2, 3, 4, 5, 6, and 7 days. Seventy flowers were randomly selected and remarked for O. longilobus and O. taihangensis receptively. The pollen was sampled every hour from 8:00-17:00 for seven consecutive days, with three replicates for each sample.
The pollen viability and stigma receptivity were detected by the I 2 -KI method at different developmental stages. Some of the pollen suspension was placed on a microscope slide using a pipette, and the pollen grains were observed using an optical microscope (10 × 40x, OLYMPUS BX51 EVOS-M7000). Five fields were randomly selected from each slide to observe the staining status of the pollen grains. The deeply/completely stained pollen grains were considered viable, according to Dafni [26].
The stigma receptivity was analyzed using the benzidine-hydrogen peroxide technique [61], wherein bubbling in the presence of hydrogen peroxide was considered to be a positive result [62,63]. The reaction was observed and recorded under an optical microscope (10 × 10x). We also recorded the time when bubbles appeared on the stigma, where their abundance around the stigma indicated viability. The absence of bubbles around the stigma indicated inactivity or sterility. The stigma receptivity was verified using the catalase activity method [25].

Outcrossing Index and Pollen/Ovule Ratio
The outcrossing index (OCI) and ratio of pollen grains to ovules (P/O) were determined to estimate the likelihood of pollination outcrossing (xenogamy) versus selfing (autogamy) based on the floral morphological characteristics [4,44,62].
For the pollen count, the anthers were collected from the synantherous stamen of mature flowers, from which three florets were collected. A total of 30 florets from ten randomly selected O. longilobus and O. taihangensis were examined. The synantherous stamen was opened with fine-tipped forceps, and the anthers of each flower were transferred to a centrifuge tube (1.5 mL) filled with 1 mL distilled water after drying. Next, a 1 µL subsample of the pollen suspension was transferred to a microscope slide using a pipette, and the pollen grains were counted using an optical stereomicroscope (10 × 10x, EVOS-M7000).
The total number of pollen grains = subsample a number of pollen grains × volume of centrifuge tube/volume of pollen suspension in the pipette. This measurement was repeated 10 times for each flower, after which the maximum and minimum values were removed before averaging. The total pollen number for each flower was recorded as P, and the number of ovules of each flower was counted and recorded as O. The pollen-ovule ratio was calculated following the method of Cruden [27].
The OCI was estimated as the sum of values from: (1) flower diameter, 0, 1, 2, 3 for opening diameters of ≤1, 1-2, 2-6, > 6 mm, respectively; (2) spatial position between anther and stigma, 0 remarked at the same height and 1 for separation from each other; (3) interval time of anther dehiscence and when the stigma was presently fertile, 0 for anther and stigma both maturing at the same time or the gynoecia initially maturing, and 1 for the stamen initially maturing [44,62]. Namely, the sum of (1) + (2) + (3) was the OCI value.
When OCI was 0, the breeding system of the species was designated as closedfertilized; when OCI = 1, the breeding system was designated as exclusively self-fertilized; when OCI = 2, it was as a parthenogenetic system for the species; when OCI = 3, the system was considered as a self-fertilized type that sometimes required pollinators; when OCI ≥ 4, the breeding system was considered as partially self-fertilized with a heterozygote pattern that required pollinators.

Seed and Germination Rates
To better characterize the mating pattern of O. longilobus and O. taihangensis, four treatments were established for the flowers: (1) control (freely pollinated flowers without any treatment), (2) freely pollinated (bagged before flowering with no emasculation), (3) geitonogamous pollinated (emasculated and bagged before pollen scatter of the florets, artificially pollinated between different flowers on the same individual), (4) artificially pollinated (bagged and outcrossed with pollen from other individuals). Each treatment involved 15 individuals. Within each individual, three capitula were randomly selected. When the seeds matured, those under each treatment were collected.
The seeds collected from each treatment after they matured from October to November from 2018 to 2019. Under the room temperature of the library, the germination experiments of the seeds were carried out. Fifty randomly selected seeds of each treatment with three replicates were placed in one culture dish for each species. Subsequently, water was regularly and quantitatively supplied for all culture dishes, and the germination changes of the seeds were recorded every day.
The germination rate was the percentage of germinated seeds to the number of seeds in the experiment, whereas the germination potential was the percentage of germinated seeds to the number of seeds in the experiment on the sixth day.

Statistical Analyses
Datasets with a normal distribution were analyzed by Student's t-test and one-way ANOVA. All analyses were performed in SPSS 20.0 (SPSS, Chicago, IL, USA) for Windows. All parameter estimates were calculated as mean ± SE (Standard Error) values. The data were plotted using Origin 2.0 software.

Conclusions
Reproductive biology is an important interdisciplinary area of plant sciences, which is essential for understanding the evolution and survival of species. For this study, we investigated the floral biology, breeding systems, and pollination ecology of two Opisthopappus species (O. taihangensis and O. longilobus). These plants exhibited a suite of floral syndromes (e.g., bisexual florets, bright white ligules, bright yellow florets, big capitulum, and distinctive fragrances) that adapted to insect-mediated cross-pollination. Comprehensive studies of the OCI, P/O ratio, and artificial pollination suggested that O. taihangensis and O. longilobus had a highly pollinator-dependent mixed mating system of cross-and self-fertilization. Outcrossing predominated in the mating system, and autonomous selfpollination in its breeding system may have provided reproductive assurance for the sustainability of the population. This allowed for successful reproduction in the harsh cliff environment of the Taihang Mountains. The reasonable number of seeds produced indicated that the species were capable of sustaining their progenies in natural populations. Although bees and flies were their effective pollinators, O. taihangensis and O. longilobus each possessed unique pollinators. As the descendant of O. longilobus, O. taihangensis with special pollinators and relatively complex floral syndromes (e.g., pink and white ligules, more numbers of ligules, and florets) exhibited improved adaptation over its ancestor. Our results not only highlighted the reproductive characteristics of an endemic species of the Taihang Mountains of North China but also provided insights into the reproductive characteristics of species that flourish in limestone mountain habitats.